NVIDIA and Microsoft have announced RTX Spark, a new Windows PC platform built around local AI performance. The platform pairs NVIDIA’s Blackwell RTX GPU technology with a Grace CPU and up to 128GB of unified memory, giving thin laptops and compact desktops enough capacity to run demanding AI, creative, and gaming workloads on-device.
The launch also includes changes to Windows aimed at supporting AI agents running locally. Microsoft is introducing new security and containment capabilities for agent-based applications, while NVIDIA is launching OpenShell to help govern how agents interact with applications, files, and AI models.
RTX Spark systems are expected this fall from Microsoft Surface, ASUS, Dell, HP, Lenovo, and MSI. Adobe is also adapting Photoshop and Premiere for the platform.
Why It Matters: Organizations are weighing how much of their AI activity should remain in the cloud and how much should move closer to users and data. The arrival of systems capable of running advanced models locally introduces new questions about where AI should live and who controls the underlying infrastructure.
- Local AI Compute: RTX Spark gives Windows PCs enough local compute for heavier AI workloads. The platform combines a Blackwell RTX GPU with 6,144 CUDA cores, fifth-generation Tensor Cores, and a 20-core Grace CPU connected through NVLink-C2C. NVIDIA says systems can reach 1 petaflop of AI performance and support up to 128GB of unified memory.
- Agent Controls: Microsoft and NVIDIA are trying to make AI agents safer to run on personal devices. Microsoft is adding Windows controls for identity, containment, policy, and management, while NVIDIA OpenShell gives users more control over what agents can access and how requests move between local and cloud models. The goal is to let agents work across files and applications without giving them unrestricted access to the whole machine. Hermes Agent and OpenClaw are among the early projects adopting these tools.
- A Unified Platform: The platform combines AI, graphics, and gaming features that were once spread across different classes of machines. RTX Spark includes CUDA, RTX, DLSS, TensorRT, OptiX, Reflex, and G-SYNC, which gives it a wide base of existing NVIDIA software support. NVIDIA says the hardware can render 90GB 3D scenes, edit 12K 4:2:2 video, generate 4K AI video, and run ray-traced games at 1440p above 100 frames per second.
- Software Support: Creative and developer applications are being tuned for the new architecture. Adobe is adapting Photoshop and Premiere to use RTX Spark’s unified memory, GPU acceleration, and TensorRT support, with NVIDIA citing up to 2x faster performance across some editing, effects, color, and generative AI tasks. Other tools named in the announcement include Blender, Blackmagic Design, CapCut, ComfyUI, OTOY, DaVinci Resolve, Affinity, MATLAB, Cursor, GitHub Copilot, Claude Code, llama.cpp, and Hugging Face frameworks.
- Windows AI Workstations: The rollout connects consumer PCs with higher-end local AI systems. Microsoft says it has optimized Windows scheduling, power management, unified memory handling, DirectX, Windows ML, and Prism emulation for RTX Spark devices. Hardware partners include Microsoft Surface, ASUS, Dell, HP, Lenovo, MSI, Acer, and GIGABYTE. NVIDIA and Microsoft also discussed DGX Station for Windows and GB300-powered workstation systems.
Go Deeper -> NVIDIA and Microsoft Reinvent Windows PCs for the Age of Personal AI – NVIDIA
Introducing a powerful new chapter for Windows PCs, accelerated by NVIDIA RTX Spark – Microsoft
Trusted insights for technology leaders
Our readers are CIOs, CTOs, and senior IT executives who rely on The National CIO Review for smart, curated takes on the trends shaping the enterprise, from GenAI to cybersecurity and beyond.
Subscribe to our 4x a week newsletter to keep up with the insights that matter.
